Experimental evaporation of Mg- and Si-rich melts: Implications for the origin and evolution of FUN CAIs

Abstract

FUN (Fractionation and Unidentified Nuclear) calcium-, aluminum-rich inclusions (CAIs) have large mass-dependent fractionations of silicon, magnesium, and oxygen isotopes (up to δ29Si ∼15‰ and δ25Mg ∼40‰), and mass-independent isotopic anomalies in many elements. To test the proposition that the mass-fractionation effects of all three isotopic systems in FUN CAIs were the result of evaporation of at least partially molten precursors, we conducted a series of experiments in which two magnesium- and silicon-rich melts (FUN1 with 53.4wt% MgO and 41.3% SiO2, and FUN2 with 32.7% MgO and 38.7% SiO2, and Al2O3 and CaO in solar proportions) were evaporated into vacuum at 1900°C for various lengths of time. The chemical and isotopic compositions of the evaporation residues were measured and compared to two of the most highly mass-fractionated FUN CAIs, Vigarano 1623-5 and Allende C1. The isotopic composition of the evaporation residues was also used to determine the kinetic isotopic fractionation factors α25,24=0.98372±0.00041 for 25Mg/24Mg and α17,16=0.9883±0.0006 for 17O/16O for residues containing >15wt% MgO, and α25,24=0.98567±0.00046 and α17,16∼0.994 for residues containing <15wt% MgO. The 29Si/28Si fractionation factor α29,28=0.9899±0.0004 was found to fit the data from the entire set of residues. Simple linear correlations were found for δ29Si, δ25Mg, and δ17O as a function of the fraction of magnesium or silicon remaining in the residues. The fact that the isotopic fractionations of magnesium, silicon and oxygen of C1 are in the same proportions as in the experimental evaporation residues suggests that the evaporation played a major role in the chemical evolution of this FUN inclusion. In the case of Vigarano 1623-5, the magnesium and oxygen isotopic fractionations are consistent with the experimental data, but fractionation of silicon isotopes relative to that of magnesium in 1623-5 is about a third less than in the experimental residues. Assuming that Allende C1 and Vigarano 1623-5 are evaporation residues that were produced in much the same way as our experimental residues (i.e., evaporation of completely molten droplets), the chemical compositions of their precursors were calculated using bulk chemical and isotopic compositions of C1 and 1623-5 together with the experimentally determined kinetic fractionation factors α25,24 and α29,28. It was found that the present chemical and isotopic compositions of C1 can be explained by evaporation of a precursor with a bulk composition close to that of a condensate from a solar composition gas. In the case of Vigarano 1623-5, however, the calculated precursor is significantly enriched in magnesium and depleted in silicon compared to plausible condensates from a solar composition gas. Among the possible reasons for such misfit could be uncertainties in bulk chemical and isotopic compositions measured in Vigarano 1623-5, or evaporation at lower temperatures from partially rather than completely molten precursors which could have different evaporation kinetics and isotopic fractionation factors.